chevron_left Back to home page
share
Case study
Vertis
Using machine learning to mitigate the risks of a high-stakes business decision
category
Services
Data Engineering, Frontend, Backend,
UX/UI Design, DevOps
language
Website
www.vertis.com
event
Year
2020
Finding the right location to live, work and play is a challenge in today’s busy world. When choosing a new job you might consider commute time, changes in cost-of-living and your quality of life, all of which are inextricably linked to the location of your employer and, in particular, your office.
 
Office location is a high-stakes investment for businesses of all scales and all industries. How can a business predict where its ideal future employees are located? How do they balance the costs and mitigate the risks?

Our client, an international industry-leader in identity and access management with over 2000 employees, wanted to expand its operations in the US but faced this same dilemma. Rather than making a subjective decision, they turned to the data and brought Acaisoft’s data science and development team on board to deliver a machine-learning based app that would provide office site recommendations. The app stretched beyond just the client to be  fully-customisable to any business's needs.

Challenges

quote
To curate vast dataset, aligning different sources and accounting for historical changes in county boundaries and names
The main objectives were to:

- Deliver data-driven recommendations for office locations - with full-customisation to the changing needs of any business.
- Develop machine learning models to forecast location characteristics - with the versatility to accept new and diverse data sources overtime.
- Ensure continuous delivery and scalability -  while giving the customer the ability to query and explore the details of a vast dataset.
- Produce an application focussed on communication and visualisation - to allow thorough understanding of the complex underlying data and models.

Data was compiled from across the US covering all aspects of the economy, population, office real-estate, and the labor pool. Covering the last 40 years, the dataset contained historical trends down to the highly-granular county level.

Here was the first major challenge of the project - to curate vast dataset, aligning different sources and accounting for historical changes in county boundaries and names. The result was a highly valuable, comprehensive dataset, encapsulating highly detailed trends for the entire United States.

Solution & Result

quote
JSON files allow the recommendations to be tailored to the customer without any changes to the code itself
An extensive dataset was pooled from across the US, covering a period of 40 years for every county in the US. The challenge of big data is to make it cohesive. Location here was key and data engineers carefully mapped the historical changes in location names and boundaries. The resulting high-quality dataset focussed on four key characteristics - economy, population, office real-estate and labor pool.

The recommendation model was built with flexibility in-mind from the inside-out. Different businesses have different needs - they rely on different sectors of the labor pool and have different cost margins. JSON files allow the recommendations to be tailored to the customer without any changes to the code itself. The model flows through a series of deep-learning autoencoders, which generate location-specific scores. A learning -to-rank model based on these scores, allows the generation of a full location recommendation report for the customer.


Creative Web & Mobile Design
vueSketch Logofigma
Hi-tech Programming & Developing
angularGOpythonjavareact
vueSketch LogofigmaangularGOpythonjavareact

Cooperation

Through the years of being in the market both our technologies and expertise have advanced and our offerings have been appreciated by our end users from the USA and the EU.
quote
Information that he is looking after the offer you can add her to Linkedin and can answer questions about the offer
photo
Laura Sanders
Marketing Director
Single Point

Contact us

We can change your idea into a real product. Based on the needs of your business, we will create a product that users will love.
Interested? Leave your email, we will contact you.
BOOK A CALL

Contact us

Thank you
Your message has been sent.
We will respond to you within 24 hours.
HOME
Oops! Something went wrong while submitting the form.
Since working with Acaisoft, our delivery velocity has increased at least 20 times. In the past months, we’ve done more developments than what had been done in the last four years in our company.
Customer Avatar
Pablo Verano
Head of Product
Scoopr
close

Contact us

Thank you
Your message has been sent.
We will respond to you within 24 hours.
HOME
Oops! Something went wrong while submitting the form.
Since working with Acaisoft, our delivery velocity has increased at least 20 times. In the past months, we’ve done more developments than what had been done in the last four years in our company.
Customer Avatar
Pablo Verano
Head of Product
Scoopr
close

Contact us

Thank you
Your message has been sent.
We will respond to you within 24 hours.
HOME
Oops! Something went wrong while submitting the form.
Since working with Acaisoft, our delivery velocity has increased at least 20 times. In the past months, we’ve done more developments than what had been done in the last four years in our company.
Customer Avatar
Pablo Verano
Head of Product
Scoopr
close